Evan Feng

Appearance

Hi, I'm Evan.

I'm currently learning LLM pretraining architecture and LLM infrastructure.

Research interests: model architecture, efficient attention, and GPU kernels.

Read my writing

Beliefs

  1. taste is important
  2. the world is predictable in intelligent ways
  3. agency harness matters a lot
  4. write things down

Recent writing

Sparse Linear Attention: Viewing SWA from the Perspective of Linear AttentionTechnicalK3's KDA Precision Problem: Why Lower-Bounded Decay Is NecessaryTechnicalFrom Heuristic Learning to Auto-Architecture SearchReflection

Notes

AlgorithmCourse Notes19 lecturesCS 336LLM Notes6 lecturesComputer NetworkingNetworking6 chapters